Does that suggest that about a half of the stars in the observable universe could not belong to any galaxy?
Not really. A key sentence in the article is "The best interpretation is that we are seeing light from stars outside of galaxies but in the same dark matter halo". So the stars are still within the dark matter halo of a galaxy, but are outside the boundary of the galaxy if the dark matter halo is not considered.
Furthermore the "intrahalo light" explanation is just one of two possible explanations according to Updated analysis of near-infrared background fluctuations which explains:
Two scenarios have been proposed to interpret the clustering excess. The first advocates the contribution from intrahalo light (IHL), i.e. relatively old stars stripped from
their parent galaxies following merging events. These stars
therefore reside in between dark matter halos and constitute a low-surface brightness haze around galaxies. The IHL
is expected to come mostly from low redshifts (1 + z <
systems (Cooray et al. 2012b; Zemcov et al. 2014).
The second scenario is instead based on the presence
of a class of early, highly obscured accreting black holes
of intermediate mass (∼ 10^4−6M⊙) at z >
∼ 13 (Yue et al.
2013b, 2014). As a suitable mechanism to produce such
objects does exist – the so called Direct Collapse Black
Holes (DCBH, for a concise overview of the problem see
Ferrara et al. 2014), and the interpretation of the super-
massive black holes observed at z = 6 seemingly requires
massive seeds (Volonteri & Bellovary 2011), such hypothesis seems particularly worth exploring.
Both scenarios successfully explain the observed clustering excess, albeit with apparently demanding requirements. In fact, if the excess is to be explained by intra-
halo light, then a large fraction of the stars at low-z must
reside outside systems that we would normally classify as
“galaxies” (Zemcov et al. 2014). On the other hand, in the
DCBH scenario the abundance of seed black holes produced
until z ∼ 13 must represent a sizeable fraction of the estimated present-day black hole abundance, as deduced from
local scaling relations (Kormendy & Ho 2013) and recently
revised by Comastri et al. (2015). However, it is important
to outline that both scenarios are not in conflict with any
known observational evidence